Machine learning mainly refers to neural networks. These are computing systems having interconnected nodes and they function similar to the neurons that we humans have in our brain. These systems use data to recognise patterns and correlations and keep classifying them to continuously learn and improve.
One of the more common examples of machine learning would be an online translator. Going back about 10 years, online translators were tried out via deep learning. In the later years they built neural networks that began to imitate the human cognitive processes and thus due to the continuous learn and improve technique the translators went on improving. Today one gets a fairly good translation on online translators compared with say five years back.
Artificial Intelligence (AI) is not a recent concept. Scientists have been exploring AI since way back in the 1950s and were quietly progressing until their efforts began gathering a strength way beyond an average person’s imagination in the late 1990s.
The momentum of the breakthroughs being accomplished from one year to the next are mind boggling. Neural networks already work just as our brain does and the process is achieving a state similar to our nervous system.
Be it facial recognition, business recommendations, forecasting of sales, market research, risk management and several other areas, it all goes down to machine learning. Investment banks and hedge funds have been using artificial intelligence to keep ahead of the stocks and currency markets and invest billions of dollars in the trades that are chiefly carried out by machines.
Diseases are being diagnosed using AI. Organizations have even implemented bots to attend to their customers online. A bot is a web robot. It is a software that is assigned automated tasks. These bots do the work of humans, only faster since the tasks are repetitive and simple.
Scientists have ceased talking about only one AI. They talk in terms of many hundreds each of which is to specialise in a complex task and many of these applications already do jobs that humans once did.
With each passing decade, humans will be prohibited from doing some of the tasks they do today. In fact, there is going to come a time when it will be illegal for humans to drive a car. This is due to the fact that self driving cars are expected to be much safer.
Not just this. Scientists are even working on colonising parts of our galaxy in time to come.
Machine Learning Engineers are among the highest earners in the world of technology. The median salary for a Machine Learning Engineer in the U.S. is as high as $146,000. This is up by nearly 350 per cent in the last five years and given the shortage, the salaries will keep on rising in the foreseeable future. The situation is the same in most of the developed world as regards the demand and remuneration. Presently, there is a gap of a few million such engineers between the supply and demand.
Given this environment, let’s have a look at some of the free online courses in Machine Learning.
This course comes to you from Stanford University. The program teaches you about the most effective machine learning techniques. While dealing with the theoretical part the course also gives you knowledge on how to use the techniques in practice.
The course covers machine learning, data mining and statistical pattern recognition. They also take you through several case studies and applications.
Machine Learning Fundamentals
The course is offered by the University of California, San Diego. This course lasting 10 weeks will require that you put in an effort of 8 to 10 hours each week.
The syllabus covers classification, regression, conditional probability estimation, generative and discriminative models, linear models and extensions to nonlinearity using kernel methods, ensemble methods and representation learning.
Machine Learning with Python – A Practical Introduction
This is a five-week course requiring an effort of 4 to 6 hours each week. The course takes you through popular algorithms including classification, regression, clustering, dimensional reduction and popular models including train/test split, root mean squared error and random forests.
PyTorch Basics for Machine Learning
This course is from IBM. The duration is five weeks and will need that you put in about four to five hours per week. The course teaches you the fundamentals of PyTorch, classic machine learning algorithms and optimising models. This is a precursor to deep learning models.
Deep Learning Specialisation
The course takes about four months to complete and they suggest an effort of five hours a week. While the medium of instruction is English the course is subtitled in English, Chinese, Arabic, French, Ukrainian, Portuguese, Vietnamese, Korean, Turkish, Spanish and Russian.
The programme covers neural networks, machine learning projects, Convolutional networks, RNNs, LSTM, Adam, Dropout, BatchNorm, and Xavier/He initialisation. They also take you through case studies from healthcare, autonomous driving, sign language reading, music generation and natural language processing.
Amazon SageMaker – Simplifying Machine Learning Application Development
This four-week course shall require that you invest about two to four hours per week. This course will teach you, an application developer, how to use Amazon SageMaker to simplify the integration of Machine Learning into your applications.
Key topics include: an overview of Machine Learning and problems it can help solve, using a Jupyter Notebook to train a model based on SageMaker’s built-in algorithms and, using SageMaker to publish the validated model.
Photos: Shutterstock / Edited by: Martina Advaney
Youth Time is hosting free webinars throughout December. Want to join us?
All your donations will be used to pay the magazine’s journalists and to support the ongoing costs of maintaining the site.